Create and log on to the Embodied Intelligence Platform
This topic describes how to create an Embodied Intelligence Platform instance in the AnalyticDB console, complete the prerequisites, and log on to the platform.
Prerequisites
An AnalyticDB for MySQL Enterprise Edition instance is created. Basic Edition instances do not support AI application features.
The instance has at least 2 ACUs (for a quick trial). If you want to use GPU training and simulation features, contact technical support to confirm the resource specifications and available regions.
Step 1: Create the Embodied Intelligence Platform
Log on to the AnalyticDB console.
In the Clusters, click the target instance.
In the left-side navigation pane, expand AI Applications and click Embodied Intelligence Platform.
Click Create Embodied Platform.
In the right-side panel, configure the following parameters, and then click OK.
Basic configuration
Parameter | Description |
Name | The platform name. The name can be up to 16 characters in length and can contain only lowercase letters, digits, and underscores (_). It must start with a letter and end with a letter or digit. Hyphens ( |
Platform service deployment resources
Parameter | Description |
Resource Specification | The computing specifications of the platform service. For a quick trial, select medium(2ACU). A total of approximately 16 specification options are available, ranging from small(1ACU) to m.8xlarge(96ACU). |
Inference service deployment resources
Configure the Head node and Worker Group for the inference service:
Parameter | Description |
Head Resource Type | The resource type of the Head node. Valid values: CPU and GPU. |
Head Resource Specification | The computing specifications of the Head node. The available options depend on the selected resource type. |
Head Disk Size | The disk capacity of the Head node. Default value: 100 GB. |
Click + Add Worker Group to add multiple Worker Groups. Each group includes the following parameters:
Parameter | Description |
Worker Group name | Automatically generated. The name can be up to 30 characters in length. |
Worker resource type | Valid values: CPU and GPU. |
Worker Resource Specification | The available options depend on the selected resource type. |
Worker Disk Space | Default value: 100 GB. |
Minimum number of workers | The minimum number of instances for auto scaling. |
Maximum number of workers | The maximum number of instances for auto scaling. |
Development and training deployment resources
Parameter | Description |
CPU ACU quota | The upper limit of CPU resources for development and training tasks. |
GPU Specification | Click + Add GPU Specification to add multiple GPU types. |
Storage configuration
Parameter | Description |
Auto-enable ADB managed data lakehouse and acceleration service | If selected, OSS storage is automatically configured for persistent storage of datasets and model files. |
The creation process takes approximately 15 minutes and involves Ray cluster and K8s Pod scheduling. During creation, the instance status is Creating. After creation is complete, the status changes to Running.
Quick trial and advanced usage
Mode | Recommended configuration | Use case |
Quick trial | Platform service: medium(2ACU), GPU disabled, inference service with minimum configuration | Evaluate platform features and familiarize yourself with the workflow |
Advanced usage | Configure GPU specifications based on training and simulation requirements. Contact technical support to confirm resources | Model training, simulation evaluation, and production deployment |
Step 2: Configure prerequisites
Configure the image repository (ACR)
Custom model development and simulation scenarios require Container Registry (ACR) to store custom images.
Log on to the Container Registry console.
Select the same region as your AnalyticDB instance.
Activate an Enterprise Edition instance of Container Registry.
Configure Internet access (optional)
Development machines are deployed in a VPC by default and cannot directly access the Internet. If you need to install Python dependencies, pull models, or access external APIs on the development machine, configure a public NAT gateway.
For detailed steps, see Enable public network access for development machines.
Step 3: Log on to the platform
On the Embodied Intelligence Platform page of the AnalyticDB console, verify that the status of the target platform is Running.
Click the link in the Platform Service URL column to open the platform login page in a new window.
Log on with the default credentials:
Username:
adminPassword:
password
After the first logon, we recommend that you immediately change the default password.
Step 4: Configure the access whitelist
For security purposes, the platform allows only IP addresses in the whitelist to access the platform by default.
On the Embodied Intelligence Platform page of the AnalyticDB console, click Configure Access Whitelist.
Enter the IP addresses or CIDR blocks that you want to allow. Separate multiple addresses with commas (,).
Click OK.